[
https://issues.apache.org/jira/browse/DAFFODIL-1539?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Brandon Sloane reassigned DAFFODIL-1539:
----------------------------------------
Assignee: (was: Brandon Sloane)
> scrutinize use of ULong.longValue
> ---------------------------------
>
> Key: DAFFODIL-1539
> URL: https://issues.apache.org/jira/browse/DAFFODIL-1539
> Project: Daffodil
> Issue Type: Improvement
> Components: Back End, Clean Ups
> Affects Versions: 2.0.0
> Reporter: Michael Beckerle
> Priority: Major
>
> We use ULong in two different ways (possibly more, but this is about these
> two)
> We use them as values coming out of data parsing or going into data
> unparsing. In this case, all 64 bits are used, and we don't care if the
> underlying stored Long integer is negative or positive.
> We also use them as bit offsets/positions into the data stream. This assures
> that they are always non-negative when created.
> However, the math that computes bit offsets calls ULong.longValue a great
> deal, then operates on the Long elements.
> This is technically incorrect. Now, one can argue that it won't matter
> because you need an offset bigger than 2^63 ... before you get a ULong where
> the underlying Long would be negative.
> But, this is the stuff of security holes. E.g., a data format can store a
> length that results in a bit length > 2^63. No such data can exist, but some
> prefix header field can be created that suggests that much data.
> And then who is to say that something won't break?
> So, the right thing here is for arithmetic on ULong to work directly (I'm not
> sure why it doesn't. There appear to be operators defined.) Or for a method
> other than ULong.longValue be used when what we are assuming we're getting
> back is a non-negative Long.
> We should scrutinize all calls to ULong.longValue and see if they can/should
> be changed to ULong.asPosition or ULong.asLength, etc. methods that express
> the intent, or the conversion can just be avoided and math done directly on
> ULong values.
--
This message was sent by Atlassian JIRA
(v7.6.14#76016)
